INDEX

Explanations

terms related to uniqueness and classification

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 24-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L24R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.24.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

æĦıæĢĿ

-0.15

éłĨ

-0.14

batim

-0.14

ibel

-0.14

utes

-0.14

 Bread

-0.14

agna

-0.14

Cul

-0.14

ajor

-0.14

Zuk

-0.14

POSITIVE LOGITS

 Slip

0.16

à¸£à¸²à¸Ĭ

0.15

0.14

 Screening

0.14

_BUF

0.14

.fun

0.14

kke

0.14

bins

0.14

 Lind

0.14

ç«¯

0.14

Activations Density 0.005%

terms related to uniqueness and classification

No Comments

No Known Activations